Big Data: Helper and game changer in laboratory medicine and genetics

Big data makes it possible to uncover connections that the human eye cannot see: A database is created by using large amounts of data from diverse patients, serving as a foundation to explore and answer medical questions. An algorithm, software, searches the database to identify patterns for making suitable predictions.

In the case of laboratory diagnostics, the shared occurrence of different values or biomarkers can indicate the presence of a specific disease. Algorithms are especially helpful when humans are confronted with large volumes of data, they are unable to process. In radiology, radiologists must frequently view hundreds of images for each patient. It is a task that is tedious and time-consuming. Algorithms could pre-screen and pre-sort these images and identify the ones that indicate anomalies. This would ease the burden on physicians.

The training of an algorithm is called "machine learning". Before an algorithm can make predictions, it needs help from humans: They must first feed it data and tell it what this data means – examples include MRI scans of tumors or pictures of the human retina. Once the algorithm provides its own results, humans must then confirm or correct them to help the algorithm improve its predictions.

Genetics is another sector that produces vast quantities of data. In the future, we could use the power of algorithms to help manage them: It is estimated that humans have about 25,500 genes, which themselves or in combination can cause diseases as genes determine the molecular processes in the body that determine whether we are healthy or sick. Due to the vast number of genes and their effects, using machine learning for pattern recognition is a great idea.

The MoSBi project at the Technical University of Munich studies the molecular mechanisms of human diseases to identify subtypes of diseases that appear similar in terms of their symptoms. However, a better distinction can be critical to make the right treatment decision and provide a more targeted therapy.

"Cancer is perhaps the most notable example of this," explain Dr. Josch Konstantin Pauling and Tim Rose in a MEDICA-tradefair.com interview. "A multitude of mutations can lead to changes in tissue, which can develop into a tumor. Doctors examine the specific changes in the tumor and adjust the patient’s treatment accordingly."

The team behind MoSBi developed the algorithm and makes it available as a free web-based application to allow researchers to analyze data from their own studies.

Another example of the benefits of big data and algorithms in laboratory medicine is single-cell sequencing. A team from the Max Delbrück Center for Molecular Medicine in the Helmholtz Association has developed ikarus, a machine learning program for just this purpose: "We developed a software that works in single-cell resolution and finds gene expression patterns to specific cell types. In case of cancer, the method found a pattern in tumor cells that is common to different types of cancer, consisting of a characteristic combination of genes," explains Dr. Altuna Alkalin in a MEDICA-tradefair.com interview.

These insights not only help facilitate the diagnosis: pathologists visually examine tissue samples to identify cancerous cells. This means they must identify the various cell types under the microscope. If the ikarus algorithm makes more information about a tumor and its immediate environment available in the future, it could also affect the medication used for targeted treatment.

No matter the problem, an algorithm will never diagnose an illness itself. Algorithms are designed to help physicians manage large amounts of patient data. In doing so, algorithms will become essential in the future as increasingly more sophisticated diagnostic methods are generating more and more data from each patient.

Admittedly, not all diseases generate big data since there are already reliable laboratory techniques in place. However, genetic disorders are still a challenge and there is much we still do not understand and for which we lack the respective treatment options. We must gain a better understanding in the long term, especially when it comes to cancer. One way is to make big data more usable and unlock its potential.